AI researcher Joshua Saxe calls out tautological reasoning in AI doom arguments

joshua_saxe · x · 2026-09-21

AI safety researcher Joshua Saxe argues that many AI doom arguments rely on a "tautological hat trick": if you assume by definition that the imagined "it" kills everyone, then "if anyone builds it, everyone dies" is trivially true. He notes the authors try to show today's LLMs will evolve into such a humanity-killer, but the book treats the premise that capricious goal-driven extinction machines would actually extinct us as carrying its own weight — a circular argument.

Original post →

More from AGI Musings

AGI Musings channel →